electrophysiology to quality of life
Klop, W.M.C.
Citation
Klop, W. M. C. (2009, April 8). Exploring the capabilities of modern cochlear implants : from electrophysiology to quality of life. Retrieved from https://hdl.handle.net/1887/13726
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the
Institutional Repository of the University of Leiden
Chapter 6
An objective method to measure electrode independence in cochlear implant patients with a dual-masker forward masking technique
W.M.C. Klop, J.H.M. Frijns, W. Soede, J.J. Briaire
Department of Otorhinolaryngology
Leiden University Medical Center, Leiden, the Netherlands
Hearing Research (In Press 2009)
Abstract
This study introduced a dual-masker forward masking technique and evaluated whether this objective method could measure electrode independency in a cochlear implant; more particularly, whether the optimal locations and number of active electrodes could be determined. This method further enabled the investigation of the efficacy of current steering, because the proposed recording method could also be described as applying a sequentially current steered masker.
The paradigm requires 5 frames involving 2 maskers and 1 probe and is referred to as the Apple Core method (MP5-AC). For each recording, both the masker and probe amplitude were varied independently, producing 3-D eCAP plots that showed the eCAP amplitude for independent variations of masker and probe amplitudes. A simple quantitative model was developed to aid interpretation of the results. Theory and model were clinically tested in 14 patients. On the basis of the model, the multi-variate, color coded plots could be subdivided into seven distinct regions, each depicting a unique relationship between the probe and the maskers. The model’s predictions supported interpretation of the results, and indicated independence for the probe electrode contacts only at lower current levels and/or at greater inter-electrode separations. The clinical results revealed a lack of selectivity in the electrode array for stimulus levels larger than 600A.
This suggests that sequential current steering is only capable of producing a single excitation area at higher current levels, or smaller electrode distances, without additional loudness correction being applied.
Thus, the MP5-AC paradigm provided insight concerning the independence of electrodes and the efficacy of current steering in clinical patients. However, its current clinical applicability is limited because measurements were adequate only in anaesthetized patients.
Introduction
Research to improve the benefit of cochlear implants focuses on the electrode- neural interface, as well as on speech coding strategies. Differences in implant performance are not only due to patient characteristics, but also due to the implant systems’ technical specifications. Although similar, commercial CI arrays differ in detail, with respect to the number of electrode contacts, distance between contacts and intracochlear position. Implant characteristics influence threshold and the amount of spread of excitation.1-6 . It is generally believed that these two parameters are related to speech perception outcomes. In studies using psychophysical data, poor electrode discrimination was negatively correlated with speech understanding.7,8 Correspondingly, several studies have shown that disabling electrode contacts with high pitch confusions, or reducing the insertion depth, positively influenced speech perception outcomes.9-12 However, deciding if, and which electrodes should be switched off remains difficult. Determining the optimum electrode density in human CI users requires time-consuming, taxing psychophysical experiments and speech understanding tests. Consequently, practical application of psychophysical methods to select electrodes is limited, especially in patients that generally do not perform well on these tasks, such as the growing population of children, multiple handicapped and prelingually deafened adults.
Recently, Dingemanse et al.13 (2006) introduced an alternative psychophysical method to assess overlap of stimulation between electrode contacts in cochlear implant patients. The method of Dingemanse et al.13 applies a psychophysical forward masking paradigm using maskers on either side of the probe. This method avoids off-site listening — the electrical analogue of off-frequency listening — and estimates the stimulation level for which there is complete overlap of the excitation patterns from adjacent electrode contacts.
In analogy with converting psychophysical spatial selectivity measurements14-18, into objective tools17,19-21, one study objective was to investigate the eCAP equivalent of the psychophysical forward masking paradigm with a dual contact masker. Potentially, the objective dual masker method is usable in all CI patients, including children, even at higher current levels. Furthermore, our method, in
keeping with objective methods generally, reduces measurement time22 and limits the confounding effect of central processing.17
A dual-masker, forward masking paradigm can also be used to investigate and improve new speech processing strategies. An acoustic signal posesses temporal and spectral properties that are critical for sound perception.23 Spectral information is encoded by the individual contacts distributed along the electrode array. Recent speech coding strategies, such as MP300024, use psychophysical masking models to improve spectral peak selection. The masking functions determine whether certain spectral peaks are actually perceptible. If not, they may be omitted from the signal without loss of information.24-26 An electrical spread of excitation model could enhance this approach. Such a model could help predict whether an electrical stimulus (on an electrode contact at a given stimulus level) elicits additional nerve fibers in the presence of stimulation from surrounding electrode contacts. Should a stimulus not introduce additional excitation, it could be safely omitted. Our objective recording method determines the excitation area between two masker electrodes on either side of the probe contact, at various stimulation levels. This would indicate the dynamic range for which the center contact can still contribute to the stimulation of the neighboring electrodes.
Assessing both sides of the probe electrode is not limited to dual masker method.
The standard single masker method is also able to obtain these results by separately making recordings on both sides of the investigated electrode.
A dual-masker forward-masking paradigm may, furthermore, shed light on virtual contacts in current steering. The use of intermediate percepts between electrode contacts, so-called virtual contacts, improves the spectral domain by enhanced placement of the excitation area along the tonotopic axis of the cochlea.27 These intermediate pitch percepts between two physical contacts are generated by splitting the current between two adjacent physical electrodes. This dual electrode stimulation technique may use, simultaneous23,27,28, or, sequential29 electrode stimulation. Simultaneous stimulation is used in the HiRes120 speech coding strategy.25,30 These studies, however, showed that the number of spectral channels that could be perceived varied considerably between patients and region
spectral channels in 57 ears implanted with a HiRes90K device. Determining the individual number of spectral channels could potentially indicate which CI users benefit most from a ‘HiRes120’, a ‘HiRes60’ or a ‘HiRes240’ paradigm.
A limitation in the version of current steering strategy currently deployed is the effect of malfunctioning electrode contacts. A defective contact would cancel two bands of channels simultaneously: one towards the base and one towards the apex. To understand the effect of malfunctioning electrode contacts, as well as to investigate a minimum number of physical contacts required within the cochlea, various studies were performed to investigate the feasibility of current steering between contacts with separations up to 4 mm.31,32
Considering dual electrode stimulation, the proposed dual contact masker recording method represents a sequential current steered masker, theoretically inducing an excitation centered around the position of the probe contact. In other words, producing the same pitch percept as that induced by the probe contact.
An excitable fiber population in the vicinity of the centrally positioned probe contact would indicate an incomplete current steering condition. The majority of the psychophysical experiments were performed at, or close by, the most comfortable stimulation level. However, in speech coding strategies, stimulation levels vary widely. Using various masker and probe levels with a dual forward masking technique might indicate at which levels current steering indeed excites the center frequency.
One might hypothesize that simultaneous use of two electrodes would lead to an increased spread of excitation. Objective spread of excitation measurements, performed with a modified forward masking paradigm, showed however, that similar spread of excitation functions are obtained with single contact and dual masker excitation functions.33 The method proposed in the present study could not only test the full excitation of the intermediate electrode contacts, but could potentially, also estimate the current level where the excitation area of the probe equals that of the masker contacts. In other words, it could demonstrate when the spread of excitation from a single center contact equals that of a current steering pair with the same induced pitch.
In the present study, an objective equivalent of the subjective psychophysical dual electrode masker selectivity paradigm is presented. The method is evaluated for a series of intra- and post-operative patient recordings, while varying multiple parameters within the same experiment. 3-D color coded plots represent the eCAP responses to varying probe and masker levels. The use of different masker and probe levels complicates data analysis, since multiple excitation configurations within the same experiment are involved (probe excitation between maskers, -overlapping with maskers, -exceeding the maskers, etc.). To differentiate the various excitation configurations, a very simple excitation model was introduced.
This strictly geometrical model predicts the theoretical excitation patterns and their boundaries, but neglects phenomena such as electrical interaction, stochasticity, anatomical details and threshold variation, effects which would blur the boundaries between the various excitation modes. To include these effects a much more sophisticated model is required.21,34-36 That degree of detail is, however, not strictly necessary in this study and would only over-complicate the analysis. Therefore, this relatively simple model is used to interpret the clinical data, represented in 3-D outcome plots. In this way the data will provide insights into masking properties, electrode independence and the efficacy of sequential current steering spanning one or more electrode contacts.
Materials and methods
The Apple Core paradigm
A paradigm was developed to measure the presence of a non-stimulated fiber population between two stimulating electrode contacts. This new approach was based on the forward masking paradigm37 and the psychophysical method to avoid off-site listening.13
Masker Probe
+
= +
-
Probe Masker
Frame A
Frame B
Frame C
A+B-C
Figure 6.1a: Selectivity measurements with the M and P on different electrodes (sMP3).
The curves on the left show the (idealized) time course of the recording (stimuli and responses) for each of three frames A, B and C. The black pulse represents the early stimulus phase, which is delivered via the masker contact; the gray pulse represents the late stimulus phase, delivered via de probe contact. The Venn diagrams in the right designate the response areas of the masker and the probe contact, which are colored black or gray respectively, if they are excited in a frame, and white if not. Frame A consists of the masker pulse artifact and the eCAP elicited by it. Frame B, called the P (probe) frame, consists of the probe pulse artifact and its associated eCAP. In frame C the masker and probe contact are stimulated sequentially. The response to the probe is smaller than in frame B, since the fibers in the region of overlap between the two response areas are in their refractory state following the masker pulse. The final result (bottom of the figure) is the response of the region of overlap. It is calculated by adding frames A and B and subtracting frame C, thereby eliminating the stimulus artifacts.
M1 P M2
Frame A on (M1 & P & M2)
Frame B (on M1 & M2)
Frame C (on P)
Frame D (on P)
Frame E (on P)
Apple Core A-B-C+D
Apple Parts E-A+B
Figure 6.1b: Masker-Probe-5-Frame-Apple Core (MP5-AC). As in (a) the black pulses represent the early stimulus phase; the gray pulses represent the late stimulus phase.
Due to a sequential masking pulse on contacts M1 and M2, a certain area of neural fibers is stimulated and put in a refractory state (frame A). Later in the same frame, probe electrode P is stimulated, which results in the activation of the neuronal population around P not affected by the maskers. This area between the two maskers defines the freedom of P and is called the Apple Core because of its shape in a Venn diagram. To remove the electrical artefact and get the eCAP response of the fibers in the Apple Core region, a masker pulse is presented on P (frames C and D), which makes its response area refractory to the probe pulse on the same electrode in frame C. The latter frame can thus be used to remove the probe artefact from frame A. To cancel out the masker artefacts, the extra frames B and D are used (Apple Core = A-B-C+D). Extra frame E is a late stimulus on the probe electrode. It is used to determine the response of the Apple Parts, which are the inverse of the Apple Core (Apple Parts = E-A+B).
Following the nomenclature introduced for the standard forward masking method (MP3)38, involving three frames (figure 6.1a), the new paradigm is called MP5-AC.
MP5-AC requires 5 recording frames involving 3 electrode contacts (2 masker electrodes M1 and M2 and 1 probe electrode P, located between M1 and M2). This determines the unstimulated area between M1 and M2 and the regions of overlap between P and the maskers. The MP5-AC paradigm is graphically demonstrated in Figure 6.1b.
Due to a sequential masking pulse on M1 and M2, a certain area of neural fibers is stimulated and put into a refractory state (frame A). Later in the same frame, probe electrode P is stimulated, activating the neuronal population not affected by the maskers. This is an area between the two maskers and, thus, the group of fibers defining the independence of P. To remove the electrical artifact, and obtain the eCAP response of the fibers excited by electrode P, a — larger — masker probe pair is presented on P (frame C). There is no response following the probe stimulus in frame C. This is due to the larger masker. Hence this frame can be used to remove the probe artefact from frame A. To cancel the masker artefacts, the extra frames B and D are required. The area of excited fibres is called the Apple Core because of its shape in a Venn diagram (Apple Core = A-B-C+D). The extra frame, E, is a late stimulus on the probe electrode. It stimulates the surrounding neural tissue. Through subtraction, it can be used to determine the response of the Apple Parts; the inverse of the Apple Core (Apple Parts = E-A+B). Hence, the Apple Core plot demonstrates the eCAP caused by stimulated neural fibres not affected by the maskers, while the Apple Part plot demonstrates the eCAP of the neural fibres affected by both probe and maskers.
We hypothesized that increasing levels of the masker stimuli would result in less independence of the center electrode, which in turn would yield a smaller amplitude of the Apple Core signal, and a correspondingly larger amplitude of the Apple Part signal.
A simplified excitation model
To aid the interpretation of the Apple Core recording results, we developed a simple model of three neighboring electrodes with equidistant separation D in a homogeneous environment (Figure 6.2).
As in the previous section, the outer masker electrodes are called and and the centre probe electrode is called. Parallel to a line through the electrodes, at distance, a band of uniformly distributed nerve fibers is positioned.In this simplified geometry, a stimulus from an electrode will result in a spherical potential distribution. For simplification, it is assumed in this 2D model that all fibers within radius from the electrode will be excited and that, is linearly related to the strength of the current.
r d
D
M1 P M2
ℓm1 ℓp ℓm2
Figure 6.2: Representation of neighboring electrodes. A two-dimensional repre- sentation of three neighboring electrodes (black squares) parallel to a band of uniformly distributed nerve fibers (small circles) at distance d. Distance d is estimated between 1 and 4 mm.44 M1PM2: Masker1-Probe-Masker2 electrodes; D: distance between electrodes (1.1.mm in the HiFocus electrode); r: radius of potential distribution; :
length of fiber band stimulated by one electrode.
For simplicity, the strictly geometrical model makes no allowance for, electrical interaction, stochasticity, anatomical details, cross-turn stimulation or threshold variation. A simple function therefore suffices to describe the length (or the number ) of excited fibers within radius , becoming excited at a certain current strength for each electrode contact. Assuming a linear relationship between the number of excited fibers and the eCAP amplitude , one can use as a measure of the expected response:
A ~ n ~ = 2√(r2 - d2) (Eq. 1)
This equation can be used to obtain the number of fibers, or length of the fiber band, that becomes excited by each electrode contact (e.g., for electrode M1:
m1 and nm1). When using the electrode contacts in a masker probe paradigm, the fibers excited by the masker contacts are in their refractory state and do not contribute to the probe stimulus response. To further simplify the model and reduce the number of variables to a minimum, the dimensionless factors and are defined as follows:
= D/d = distance ratio (Eq. 1a)
= /d = excitation factor (Eq. 1b)
With these definitions, Eq. 1 reduces to:
= 2√(r2/d2 -1) (Eq. 2)
Depending on the probe and masker level, seven patterns of excitation can be distinguished, as graphically shown in Fig. 6.3 (probe in red, masker in blue).
Figure 6.3: Effect of stimulus levels on excitation patterns. Graphical two- dimensional representation of seven (I-VII) excitation patterns of a probe with two neighboring masker electrodes with variable current levels (red: probe;
blue: masker; green: overlap). I-II-III represent non-overlapping regions of excitation, in IV the masker regions don’t overlap, but they overlap partially with the probe region. In V and VI the probe region comprises the non-overlapping (V) or overlapping (VI) masker regions. In VII the masker regions overlap, leaving no region to be excited by the probe contact between them.
These patterns of excitation correspond with seven different regions (I to VII) that can be identified in Figures 6.4a and b. The plots demonstrate increasing probe levels on the horizontal axis and increasing masker levels on the vertical
Figure 6.4: Model representation of Apple Core and Apple Parts measurement
The modeled representation of isopotentials is shown as a function of number of excited fibers for growing probe levels (horizontal axis) and growing masker levels (vertical axis). Blue represents a few, red represents a large number of excited fibers.
The first three-dimensional plot (A) represents Apple Core measurements; the second three-dimensional plot (B) represents Apple Parts measurements. Seven regions (I-VII) are distinguished, as already illustrated in figure 6.3.
The two bottom two-dimensional plots (C and D) show cross-sections of the Apple Core plot (A). Plot (C) shows stable probe level with growing masker levels ( and ); Plot
Region I (Eq. 3) represents the subthreshold state of the probe when rp < d and no response is obtained.
Aresp = 0 with p 0 (Eq. 3)
The second area (Region II (Eq. 4)), is formed by a suprathreshold probe while both masker amplitudes are still below threshold (rm < d, m < 0) and the method reduces to an MP3 response on single electrode P:
Aresp =Ap with m 0 (Eq.4)
Region III (Eq. 5) has the same excitation as Region II, with the exception that the maskers are above threshold. However, there is no overlap between the excitation areas of the maskers and probe.
Aresp = Ap with 2 -½( m1 + m2)> 0 ∧ p < 2 -½( m1 + m2) (Eq. 5)
Because the eCAP in this case is determined only by the probe level, vertical isopotentials can be seen.
Region IV (Eq. 6) shows overlap between the areas of excitation of probe and masker contacts. The maskers block the edges of the probe excitation area. The response leads to horizontal isopotentials in the case of fixed-level maskers and increasing probe levels, as a larger probe cannot excite more fibers. In fact, this is the situation that gave the Apple Core method its name. This condition is described by:
Aresp~ 2 -½(m1+m2) with 2-½(m1+m2) > 0 ∧ 2-½(m1+m2)
< p < 2+½(m1+ m2) (Eq. 6)
Region V (Eq. 7) is characterized by a very broad region of excitation of the probe.
The response is a superposition of the response region between the maskers and two additional regions outside the maskers
Aresp ~ p -( m1 + m2) with 2 -½( m1 + m2) > 0 ∧ p < 2 +½( m1 + m2)
(Eq. 7)
As the outer regions steadily grow and the gap between the maskers decreases
in Figure 6.4a. In the remaining two regions (Regions VI and VII), the area between the two maskers has been reduced to zero. This will reduce the aforementioned response to only the outer areas (Region VI (Eq. 8) and results in a steeper slope of the isopotentials.
Aresp ~ p - (2 + ½(m1 + m2) with 2 -½( m1 + m2) 0 ∧ p > 2 +½( m1 + m2) (Eq. 8)
Region VII (Eq.9) is characterized by large maskers outstripping the area excited by the probe. The response is equal to zero.
Aresp = 0 with 2 -½(m1 + m2) 0 ∧ p <2 + ½(m1 + m2) (Eq. 9)
The inverse of Apple Core is the Apple Part. It describes the areas of overlap between the regions excited by the maskers and probe, which have the shape of Apple Parts in region IV. When the size of the Apple Core decreases, the Apple Parts grow and vice versa. The plot of the Apple Parts (Figure 6.4b) shows isopotentials that can be divided into the same seven regions as in the Apple Core plot (Figure 6.4a). Region I represents the subthreshold state of the probe when rp < d and no response is measured. The second area (Region II) is formed by a suprathreshold probe with both masker amplitudes still below threshold (rm < d).
Although the probe is measurable and leads to a response in the Apple Core plot (Figure 6.4a, Region II), there is no overlap and therefore, no response depicted in the Apple Part plot. The third region (Region III) has the same excitation as described in Region II, with the exception that the masker and probe are above threshold. However, there is no overlap between the excitation area of the maskers and probe and no response is measurable. When the excitation regions of the masker and probe start overlapping, the parts start to grow. This is the next region (Region IV). In contrast to Region IV in Fig. 6.4a, where we find horizontal isopotentials, the Apple Part isopotentials are curved, because the overlap in this area is influenced by both the changing probe and masker levels. The neighboring region on the right is Region V: where the probe level exceeds both masker areas.
In this situation the level of the maskers is the determining factor for the size of the Apple Part, so at fixed-level maskers horizontal isopotentials can be found.
This is an important difference compared to the Apple Core in Figure 6.4a, where the isopotentials in Region V show a positive slope. The two regions in the top
of the figure show results in which the area between the two maskers has been reduced to zero. The probe reaches over the masked areas in Region VI, leading to horizontal isopotentials determined by the maskers. In Region VII, the probe increases while the large maskers still exceed the probe region. Only the probe level determines the Apple Parts, leading to vertical isopotentials. They increase with higher probe levels.
To clarify the contents of the plot, Figures 6.4c and d represent specific situations from Figures 6.4a and b with stable probe or masker. Figure 6.4c demonstrates the change in the area of excited fibers at growing masker levels for one stable probe level. In this plot, 1 arbitrary current unit (CU) represents the threshold equivalent. The vertical axis shows the area (~ number) of excited fibers while the horizontal axis displays the masker level. The arrow () in Figure 6.4a indicates a stable probe of 2.8 CU with increasing maskers. This situation is shown in Figure 6.4c, represented by line (). For masker levels below 1.0 CU, the area of excited fibers is stable and is only determined by the probe. This is the situation of Region II in Figure 6.4a. When the masker increases from 1.0 to 2.0 CU, the maskers block the edges of the probe-excitation area causing a negative slope in Figure 6.4c. This is the Apple Core. At large masker levels (exceeding 2.0 CU), the area between the maskers is reduced to zero. The probe cannot excite fibers in this blocked area and the number of excited fibers is zero (Region VII, (Figure 6.4a)).
At a probe level of 4.8 CU (arrow (Figure 6.4a)), the slope of curve in Figure 6.4c is comparable to curve . In this situation, as long as maskers remain below 1.0 CU, the probe only excites the fibres that do no overlap with the small maskers (Region II). From masker level 1.0 CU, a downward slope results from the decreasing gap between the maskers (Region V). From masker level 2.0 CU, the flattening of the downward slope is caused by the closed gap between the maskers and stimulation of the outer areas due to the large probe overlapping both maskers (Region VI). For masker levels exceeding 3.2 CU, no fibers remain within the excitation range of the probe (Region VII).
Comparably, Figure 6.4d shows curves that represent responses with stable masker levels. The horizontal axis displays the probe level and the vertical axis presents the area (~ number) of excited fibers. There is a threshold of 1.0 CU (see
due to a growing probe. The growing probe is not influenced by the subthreshold maskers and is represented by Region II in Figure 6.4a. Initially, a steep slope (probe level between 1.0 and 1.2 CU) can be identified. The shape of this curve resembles that of the origin of the curve, which is the input/output (I/O) curve.
Line , with a stable masker level of 1.5 CU and a probe threshold of 1.0 CU, has an increasing slope from 1.0 to 1.2 CU. Suprathreshold maskers and probe, without overlap, cause this small area (region III in Figure 6.4a). The end of this region and beginning of region IV is marked by a sharp angle at a probe level of 1.2 CU. This is the Apple Core region. Because maskers are fixed-level, the area of fibers between the maskers excited by the probe, is stable. When the probe exceeds a level of 3.2 CU, fibers lateral to the masker region are excited, thus increasing the response amplitude. This is Region V in Figure 6.4a. Line in Figure 6.4d represents the situation of a masker level of 2.4 CU. The maskers’ patterns have no space between them, and small probe levels do not excite the fibers (Regions I and VII (Figure 6.4a)). Only when the probe has increased to a level of 4.0 CU, does it bypass the regions excited by the maskers. At that point, new fibers become excited and the slope increases. This is represented by Region VI in Figure 6.4a.
The simplified model provides a means to predict the influence of several parameters of the MP5-AC method. In Figures 6.5a-f, simulations of Apple Core and Apple Part measurements are shown with decreasing inter-electrode distance D (and hence ). The seven regions differ in the amount they are influenced by the changes in D (cq. ). Regions I and II are independent of , while regions IV and VI are highly influenced because is part of the equation for the response amplitude. For equations 5–9 (regions III–VII), is part of their boundary conditions. Hence, it follows that changing will lead to a different excitation profile and accompanying plots, as demonstrated in Figure 6.5.
Figure 6.5: Effect of changing the distance ratio on excited fibers in our model. The left panel shows decreasing for Apple Core measurements, the right panel for Apple Parts measurements.
At electrode distance = 0.9, all regions are present. For = 0.6, region III becomes very small, while regions IV and V are also considerably smaller. The borders between the Regions VI-VII and IV-V move upward. These borders represent
in size. Based on this simplified model, we may conclude that reducing will lead to a reduced independency of masker and probe (Region III) and diminishing independency between maskers (Regions IV and V). Decreasing the relative inter- electrode distance to 0.3 shows unaltered Regions I and II, but Regions III, IV and V have disappeared due to lack of independency of electrodes. Region VI, defined by the probe exceeding the maskers, has an increased size. Region VII, defined by maskers exceeding the probe, decreases because the maskers have an increasing overlap.
Clinical testing
To test our theory and model, we conducted a series of clinical experiments.
All tests were performed with the CII cochlear implant at the Leiden University Medical Center (LUMC). A detailed description of the 16-channel cochlear implant and eCAP recording facilities can be found in a previous paper.39 In short, the Clarion CII has an on-chip differential amplifier with multiplexed inputs, which recovers from an overload condition due to a stimulus artefact within 20μs:
thereby eliminating the need to switch off the inputs during stimulus delivery.
The responses were captured with a 9-bit (8-bits plus sign) analogue-to-digital converter operating at a sampling rate of 56kHz. Advanced Bionics Corporation supplied the software. The Visual Basic NRI software was used, enhanced with LUMC extensions. For all NRI recordings, monopolar anodic-leading biphasic pulses with a 32μs phase duration were used and an average of 32 sweeps (+/- 2 sweeps per second) of each stimulus presentation. The masker-probe interval was set at 500μs; the amplifier gain was 300. In this study, we used varying stimulus and masking electrodes as described in the Results section. Responses were typically recorded three electrode contacts apical to the probe electrode. The stimulus —, masking — and recording electrodes were short-circuited during approximately 190μs immediately preceding the stimulus onset to discharge them before each NRI sweep. Next, the NRI recording was digitally blanked until 200μs after stimulus onset to prevent any residual artefact from disturbing further post processing, which consisted of zero-phase shift filtering based on a fourth-order Butterworth low-pass filter with a cut-off frequency of 6kHz.
To reduce measuring time, we modified the MP5-AC paradigm slightly to avoid duplicate recordings of identical stimulus conditions. At every probe level,
increasing masker levels are used. As can be seen in Figure 6.1b, frame B is independent from the probe stimulus, as is frame D from the masker stimulus.
Therefore, in every new cycle, these frames can be reused. This may reduce measuring time by almost 50%. To reduce the amount of noise in the recordings, the reused frames B and D were measured with 64 instead of 32 averages.
The data were analyzed in Matlab version 5.3 (The MathWorks, Inc., Natick, MA), using the same plots as in the model.
All participants in our study volunteered, and were postlingually deafened patients implanted with the Clarion CII. All surgical procedures were uneventful.
Seven recordings were measured during the implant procedure, in the operation theatre under general anaesthesia, and seven recordings took place in an outpatient setting in awake subjects. Patients under general anaesthesia were generally stimulated with higher current levels. Behavioral T-levels were measured for all subjects. In the outpatient setting, this took place before the MP5-AC measurement; for subjects who had their MP5-AC measurement during general anesthesia, we performed the T-level determination within three months after implantation.
The Institutional Review Board of the Leiden University Medical Center granted approval under CME number P02.106.
Results
Figure 6.6 illustrates MP5-AC measurements in one of our anesthetized patients.
In these recordings, the probe stimulus was presented on electrode number 8 (P8) and the masking stimuli on electrodes 7 (M7) and 9 (M9). The horizontal axis shows increasing current level of the probe stimulus. The vertical axis shows increasing current level of the masking stimuli. For this experiment we used increasing current steps of 50μA for both masker and probe. Behavioral T-levels of the whole array were measured 6 weeks after implantation at a pulse width of 75μs per phase (30-64μA, average 45μA).
Figure 6.6: Results from clinical experiments. (a) Clinical Apple Core measurements for electrodes M7P8M9. (b) Clinical Apple Parts measurements for electrodes M7P8M9 Figure 6.6 is divided in two plots representing the same measurement. The top plot (Figure 6.6a) represents Apple Core measurements, while the bottom plot (Figure 6.6b) is the inverse of the top plot and represents Apple Part measurements.
Data are represented as isopotentials with no response in dark blue and a large response in red. To explore the clinical measurements in Figure 6.6a and b, and to identify the different regions, we used the model results in Figure 6.4.
In Figure 6.6a the border of Region I (probe subthreshold area) can be identified at a probe level of 314μA. Region II (masker subthreshold) can be found up to a masker level of 271μA. This region contains the vertical isopotentials representing only the probe level not disturbed by the maskers. These regions are bounded at the same probe (314μA) and masker (271μA) level in Figure 6.6b.
Region II can be distinguished from region V by the sharp angle of the isopotentials at a masker level of 271μA. The oblique shape of the isopotentials in region V is due to a superposition of the response region between the maskers and two additional regions outside the maskers
A continuous border estimated at a masker level of 271μA runs between region II and regions III, IV and V. At first sight, the Apple Core plot (Figure 6.6a) does not reveal the Regions III and IV. However, the Apple Part plot (Figure 6.6b) contains an indication of small Regions III and IV between probe level 314μA (threshold) and 543μA. Region III is bordered by the masker threshold (271μA), probe threshold (314μA) and region IV. The curved isopotentials of Region IV can be distinguished from the horizontal isopotentials in region V and the vertical isopotentials in region VII. Therefore, although not very clear, Regions III and IV must be represented in the Apple Core plot as well, between probe levels 314μA and 543μA. With increasing masker level in figure 6.6a, the end of region V is recognized as an increased slope at the border of Regions V and VI (masker at 393μA). In the Apple Part plot, the isopotentials are horizontal (i.e., at fixed- masker level) because the response is due only to the masker (probe exceeds maskers; masker regions do not overlap). As demonstrated, the isopotentials in the Apple Part plot remain horizontal at higher masker levels and Region V cannot be separated from Region VI, while the Apple Core plot shows the aforementioned sudden increase of the slope in the course of the isopotentials. This is due to growing masker levels that finally lead to overlap of the regions excited by the maskers. The area of the probe, which exceeds that of the maskers, now, provides the response. Region VI represents a large area in Figures 6.6a and b: There is no space between the maskers and the response is provided only by the excited areas
compared to Region V.
Region VII in the Apple Core measurement (left of Region VI in Figure 6.6a) shows no response because there is no gap between the large maskers, and the maskers exceed the probe. In the Apple Part plot (Figure 6.6b), Region VII shows vertical isopotentials, as expected: the parts do not increase with larger maskers and fixed-probe.
In summary, these illustrative plots demonstrate an NRI probe threshold at P8 of 314μA and an NRI masker threshold at M7 and M9 of 271μA. Furthermore, they reveal the existence of a gap between M7 and M9, seen through the clear presence of region V; M7 and M9 are independent from each other for masker current levels below 393μA. Finally, unstimulated nerve fibres remain as long as the probe stimulus levels are below 543μA, with masker levels below 393μA.
Figures 6.7a-f show a range of measurements (under general anaesthesia), obtained while decreasing the gap between probe and masking electrodes. This figure is comparable to the theoretical experiment presented in Figure 6.5. The three plots on the left show the result of Apple Core measurements, where the inter-masker gap is decreased from 6 to 4 to 2 inter-electrode distances (Figures 6.7a, c and e). The three plots on the right show the corresponding Apple Part measurements. The maskers were consecutively set on electrodes M5M11, M6M10 and M7M9 and the probe was set on electrode P8. The behavioral T-levels were measured 4 weeks after implantation during the first hook-up, using a pulse width of 21.6μs, and were found between 195 and 300μA with an average of 241μA.
Figure 6.7: Effect of electrode spacing in clinical Apple Core and Apple Parts measurements; comparisons with sMP3. (a-f). Apple Core (left panel) and Apple Parts (right panel) measurements in one patient. Probe electrode remains on P8;
masker distance is decreased. Response is measured on electrode 4. (g) Selectivity measurements at probe electrode 8 for different current levels. Responses recorded on electrodes 6 and 10, yielding two curves for each current level. Horizontal axis shows electrode number: vertical axis shows eCAP level. (h) Same sMP3 measurements normalized whereby the eCAP amplitude at electrode 8 is set on 1. Horizontal dashed
Figures 6.7a and b represent Apple Core and Part measurements for M5P8M11. As in the simple model in figure 6.3, seven regions can be identified. Nevertheless, there are significant differences and in this clinical case, interpretation of the borders is problematic. Region I ends at a probe threshold of 472μA, and Region II at a masker threshold of 435μA. At a probe level between 472μA and 585μA, and a masker level between 435μA and 569μA, Region III is discernible in the Apple Part plot. This is the region without overlap of the three electrodes. The detection of the borders of Region IV can be simplified by identifying Region V first. Region V borders Region IV to its right side and is characterized in the Apple Part by horizontal isopotentials. The upper border of Region V is at a masker level of 569μA and can be identified by the larger gradient of the isopotentials curves in Region VI in the Apple Core plot. Hence, Region IV is bounded at a probe level of 889μA and a masker level of 569μA. Regions VI and VII are considerably smaller compared to the following plots, consistent with model-based Figure 6.5.
If we decrease the gap by one electrode (M6P8M10; Figure 6.7c and d), the Regions I and II are comparable to the former plot. The probe threshold is 472μA;
the masker threshold equals 439μA. Region III is barely discernible, but Regions IV and V are probably present, as indicated by the rounding of the isopotentials in the Apple Part plot at a probe level between 472μA and 531μA. This indicates independence between the masker electrodes. Region VI starts at a masker level of 558μA, which is interpreted as the end of the independence between the maskers. Region VII appears on the left border of Region VI. For M7P8M9 (Figures 6.7e and f), Regions I and II are bounded by a probe threshold of 472μA and a masker threshold of 435μA. Regions III, IV and V are not discernible, but Regions VI and VII can be easily identified. According to this plot, no masker level exists for which there is independence for P8 between the masker electrode contacts.
One of the main observations in Figure 6.7 is the rounding of the isopotential curves, seen at probe and masker levels just above 450μA in the Apple Part plots with increasing masker distance. This is important because the Apple Core (Region IV) becomes discernible when the angle of the isopotential in the Apple Parts plot is no longer sharp. The plots from this patient lead us to conclude that free dynamic space exists between probe and maskers at a distance of three electrodes (M5P8M11), but only for low current levels. Another important aspect
expected), but varying masker thresholds on different electrode contacts.
To compare the new MP5-AC method with the single-masker method for the Masker and Probe on different electrodes (sMP3), we also performed selectivity measurements in this patient (Figures 6.7g and h). Selectivity was measured for probe P8 (with equal probe and masker amplitude). For each current level, the masker electrode was consecutively changed along the array from 1 (apical end) up to 16 (basal end). The responses were recorded on electrodes 6and 10, yielding two curves for each current level in Figure 6.7g. As expected, the curves show larger eCAP amplitudes with higher current levels. A peak is visible around electrode 8, the position of the probe. The spread of neural excitation in these plots was measured as described by Cohen et al.17, by measuring the widths of the eCAP functions at 50% of the normalized peak amplitude. The normalized functions are shown in Figure 6.7h, where the amplitude at electrode contact 8 is set at 1.
The curves broaden at higher current strengths, which imply less selectivity, as was expected for such high stimulus levels. For example, the blue lines representing a current level of 600μA at P8 have a spread of excitation (measured at 50% of the normalized peak) ranging from electrode 5 to 10 and 4 to 10, respectively.
The same measurement performed at 1200μA (orange curves) leads to a spread of the excitation ranging from electrode 3 to 13. Returning to the MP5-AC plot for M5P8M11 (Figure 6.7a), we find Regions III and IV at probe stimulus levels between 472μA and 889μA. (Region III contains the free dynamic range of the probe; Region IV contains the growing apple parts.) The border between Regions III and IV contains an important feature: it marks the situation with full independence and no gap, and therefore optimal selectivity for the electrodes. This border appears to be present at a probe level between 472 and 585μA and a masker level between 435 and 569μA. Hence, for this electrode contact combination, selectivity as measured with MP5-AC exists when the probe and masker levels are less than ~ 600μA. Figure 6.7(a), for example, also indicates that the response of a probe stimulus of 1200μA can be found in Regions II, V and VI, depending on the masker level. This means in this case that a probe current of 1200μA will overlap the area of suprathreshold masker electrodes and that there is no independence between them. In line with this, the sMP3 curves for 1200μA (Figure 6.7h) are almost flat between contacts 5 and 11.
A main goal of this study was to find a way to measure independency of each electrode contact of the array. Hence, the MP5-AC measurements were per- formed at different probe locations along the array for clinical patients. Those measurements revealed variable outcomes. For instance, we found that, within a patient, thresholds and regions differ along the array. Moreover, in some patients we had reproducible Apple Core plots in the basal part of the electrode, while the apical measurements resulted in a noisy, low-amplitude signal.
Measurements in eight patients (six awake and two anesthetized) out of fourteen did not appear to reveal any interpretable plots. This may be explained in two ways: First, the regions of interest (Region III, IV and V) are generally found at low current levels and thus easily disturbed by noise. Second, in awake patients comfort levels were generally exceeded with stimulation levels above 800μA.
As a result, Regions V, VI and VII were not properly defined, which hampers the interpretation of the whole plot.
Discussion
The MP5-AC, also called the Apple Core method was developed to objectively determine the independence of each electrode contact in a multi-electrode array. This paradigm requires the recording of 5 frames involving 2 masker electrodes and 1 probe electrode to determine this eCAP measure of spatial selectivity. A simple quantitative model was developed to aid interpretation of the paradigm. Theory and model were clinically tested in 14 patients. The results from these measurements were presented in plots showing the eCAP amplitude for independent variations of masker and probe amplitudes. These plots could be subdivided into seven distinct regions, each depicting a unique relationship between the probe and the maskers. Using our simplified excitation model, we were able to interpret the clinical findings and demonstrated that the Apple Core method is feasible. Compared to a single masker method, with multi variate maskers, the MP5-AC is less time consuming and provides direct information on electrode independence. The outcomes of our study are in line with the findings of its psychophysical equivalent13 as they demonstrated that with growing dual
contact maskers an upper limit in probe detection was found.
In the introduction we have argued that objective methods in CI research have considerable advantages. However, psychophysical methods allow us to measure at lower current levels, which is important when determining electrode independence. On the other hand, the objective method (under general anesthesia) allows for higher stimulus conditions. Loudness limitations prevented these conditions in the psychophysical experiments. Although recording of the MP5-AC in our study was time consuming because of the multiple parameter variation, it required less time than the time required to measure the psychophysical equivalent.
The model and patient recordings illustrated independence of electrode contacts at lower current levels and at wider inter-electrode separations (Figure 6.5).
The various predicted regions could be identified in the clinical data, but note that this represents the authors’ interpretation of the plot — based on the model predictions. Region I with sub threshold probe is of minor interest, only indicating the level of the threshold. Region II is a series of identical I/O curves, due to the sub threshold maskers. Regions III, IV and V do have clinical meaning. For instance, interpreting the sequential dual masker as a current steering pulse29 in a speech coding strategy would require the excitation of the full neural population between the maskers to produce a percept equal to the one produced by the intermediate physical contact. The existence of Regions II, III, IV and V shows that at these masker levels sequential current steering does not function correctly.
At higher masker levels, as represented by Regions VI and VII, it appears that the gap between the maskers is closed and a central percept is feasible. This would indicate that the patient presented in figure 6.6 has only full current steering at current levels above 393 A in the presence of a separating contact. The patient depicted in figure 6.7 has no, or only very small, Regions III, IV and V with one separating contact and could have current steering at all levels. When the contact separation increases, the intermediate contact acquires more independence and sequential current steering will only induce a single excitation area at high levels. Experiments by Saoji and Litvak33, demonstrating the possibility of current steering spanning one or more electrode contacts, were also performed
sequential dual electrodes stimulation indeed induced two separate excitation areas. These areas could be fused by applying loudness corrections. Investigating current steering further with the MP5-AC method can provide new insights. For example, the effectiveness of simultaneous current steering could be investigated with a similar method, but then the maskers should be presented simultaneously.
If we analyse our data with the electrical equivalent of the psychophysical masking models in mind (used to improve spectral peak selection), the stimulus levels of Regions II, III, IV and V still allow independent stimulation of neighboring masking electrodes. At current levels of the neighboring contacts (region VI and VII) the central electrode contact could just as well be omitted. The boundary between these regions shifts to higher stimulus levels with increasing electrode separation, indicating less masking at greater contact distances. A limitation of the two simultaneous maskers as a masking function in the MP5-AC method is that asymmetries in interaction will not be evident.
Similar experiments could be set up as extensions of the normal spatial selectivity forward masking paradigm (sMP3). For example, it is possible to predefine an electrode configuration and obtain consecutive recordings for different masker and probe stimulus levels. This could give direct insight into the masking properties between electrode contacts, neglecting the electrode interaction.
As with the Apple Part and Apple Core plots, the inverse response of the sMP3 could be estimated through the subtraction of an additional recording. Such data might give additional insights into the recorded responses as was the case for the MP5-Apple Part (personal communication L.T. Cohen). Furthermore, the simple computer model like the one presented here could contribute to the analysis of different regions in a multi- variate sMP3 plot. For instance, a multi- variate analysis of the sMP3 would produce 6 separate regions that are symmetric along the diagonal (if the spread of excitation of both the masker and the probe are equal). These simple models are convenient tools, but are inadequate for in-depth analysis of the results, as they lack interaction, stochastic and refractory properties. To evaluate the data on a patient to patient basis, and to extrapolate the outcomes to other neural properties, more sophisticated models should be used.21, 41-43
In the majority of patients, the MP5-AC technique did not provide clear information on the independence of electrode contacts. Only in 6 out of 14 cases were plots interpretable. This greatly reduces the applicability of the current method for patient to patient analysis, and, for instance, limits its usefulness in electrode contact selection for deactivation. Another limitation is the need to identify the lower level regions (I-IV) before determining Regions V, VI and VII. Nevertheless, the method enables the identification of independency of electrode contacts, while the sMP3 method indicates spread of excitation as a function of contacts along the array. Furthermore, in some patients, at lower stimulus levels, the MP5-AC method shows (figure 6.6) that the centre electrode is able to excite an independent fibre population in the presence of two masker electrode contacts.
In conclusion, the MP5-AC paradigm provides insight into the independence of electrode contacts and the efficacy of current steering. The main observation is the lack of independency in neighboring contacts at stimulation levels above 600μA. Next to variation in position, variation in levels of masker and probe provides new insights in the excitation pattern and the independence of contacts from threshold to maximum comfortable level. Additionally, the use of both the masked and unmasked response of the probe excitation area (called the apple parts and apple core respectively) has produced valuable information. However, with present recording systems the method is an unsuitable tool in patient populations. The development of new hardware with increased averaging speed and reduced noise levels would enable it to produce reliable results within an acceptable time window.
References
Marrinan MS, Roland JT Jr, Reitzen SD, Waltzman SB, Cohen LT, Cohen NL. Degree of 1.
modiolar coiling, electrical thresholds, and speech perception after cochlear implantation.
Otol Neurotol 2004; 25(3):290-4.
Cohen LT, Lenarz T, Battmer RD, Bender von Saebelkampf C, Busby PA, Cowan RS. A 2.
psychophysical forward masking comparison of longitudinal spread of neural excitation in the Contour and straight Nucleus electrode arrays. Int J Audiol 2005; 44 (10): 559-66.
Cohen LT, Saunders E, Knight MR, Cowan RS. Psychophysical measures in patients fitted with 3.
Contour and straight Nucleus electrode arrays. Hear Res 2006; 212 (1-2): 160-75.
Polak M, Hodges A, Balkany T. ECAP, ESR and subjective levels for two different nucleus 24 4.
electrode arrays. Otol Neurotol 2005; 26(4):639-45.
Hughes ML, Abbas PJ. Electrophysiologic channel interaction, electrode pitch ranking, and 5.
behavioral threshold in straight versus perimodiolar cochlear implant electrode arrays. J Acoust Soc Am 2006; 119(3):1538-47.
Fitzgerald MB, Shapiro WH, McDonald PD, Neuburge HS, Ashburn-Reed S, Immerman S, 6.
Jethanamest D, Roland JT, Svirsky MA. The effect of perimodiolar placement on speech perception and frequency discrimination by cochlear implant users. Acta Otolaryngol 2007;
127(4):378-83.
Zwolan TA, Collins LM, Wakefield GH. Electrode discrimination and speech recognition 7.
in postlingually deafened adult cochlear implant subjects. J Acoust Soc Am 1997;
102(6):3673-85.
Donaldson GS, Nelson DA. Place-pitch sensitivity and its relation to consonant recognition 8.
by cochlear implant listeners using the MPEAK and SPEAK speech processing strategies. J Acoust Soc Am 2000; 107(3):1645-58.
Frijns JH, Klop WM, Bonnet RM, Briaire JJ. Optimizing the number of electrodes with high- 9.
rate stimulation of the clarion CII cochlear implant. Acta Otolaryngol 2003; 123(2):138-42.
Gani M, Valentini G, Sigrist A, Kós MI, Boëx C. Implications of deep electrode insertion on 10.
cochlear implant fitting. J Assoc Res Otolaryngol 2007; 8(1):69-83.
Kos MI, Boex C, Guyot JP, Pelizzone M. Partial withdrawal of deeply inserted cochlear 11.
electrodes: observations of two patients. Eur Arch Otorhinolaryngol 2007; 264(11):1369-72.
Finley CC, Holden TA, Holden LK, Whiting BR, Chole RA, Neely GJ, Hullar TE, Skinner MW.
12.
Role of electrode placement as a contributor to variability in cochlear implant outcomes.
Otol Neurotol 2008; 29(7):920-8.
Dingemanse JG, Frijns JH, Briaire JJ. Psychophysical assessment of spatial spread of 13.
excitation in electrical hearing with single and dual electrode contact maskers. Ear Hear 2006; 27(6):645-57.
Lim HH, Tong YC, Clark GM. Forward masking patterns produced by intracochlear electrical 14.
stimulation of one and two electrode pairs in the human cochlea. J Acoust Soc Am 1989;
86(3):971-80.
Chatterjee M, Shannon RV. Forward masked excitation patterns in multielectrode electrical 15.
stimulation. J Acoust Soc Am 1998; 103(5 Pt 1):2565-72.
Boëx C, Kós MI, Pelizzone M. Forward masking in different cochlear implant systems. J 16.
Cohen LT, Richardson LM, Saunders E, Cowan RS. Spatial spread of neural excitation in 17.
cochlear implant recipients: comparison of improved ECAP method and psychophysical forward masking. Hear Res 2003; 179(1-2):72-87.
Chatterjee M, Galvin JJ 3
18. rd, Fu QJ, Shannon RV. Effects of stimulation mode, level and location on forward-masked excitation patterns in cochlear implant patients. J Assoc Res Otolaryngol 2006; 7(1):15-25.
Peeters S, Van Immerseel L, Zarowski A, Houben V, Govaerts P, Offeciers E. New 19.
developments in cochlear implants. Acta Otorhinolaryngol Belg 1998; 52(2):115-27.
Cohen LT, O’Leary SJ, Saunders E, Knight M, Cowan RSC. Modeling methods tailored to 20.
human psychophysical and ECAP data: practical applications to sound processing. 2001 Conf. on Implantable Auditory Prostheses, Asilomar, Pacific Grove, California, USA, 19–24 August 2001.
Frijns JH, Briaire JJ, Grote JJ. The importance of human cochlear anatomy for the results 21.
of modiolus-hugging multichannel cochlear implants. Otol Neurotol 2001;.22(3):340-9.
Hughes ML, Stille LJ. Psychophysical versus physiological spatial forward masking and the 22.
relation to speech perception in cochlear implants. Ear Hear 2008; 29(3):435-52.
Firszt JB, Koch DB, Downing M, Litvak L. Current steering creates additional pitch percepts 23.
in adult cochlear implant recipients. Otol Neurotol 2007; 28(5):629-36.
Nogueira W, Büchner A, Lenarz T, Edler B. A psychoacoustic “NofM”-type speech coding 24.
strategy for cochlear implants, EURASIP. J Applied Signal Processing 2005; 1.
Büchner A, Nogueira W, Edler B, Battmer RD, Lenarz T. Results from a psychoacoustic 25.
model-based strategy for the nucleus-24 and freedom cochlear implants. Otol Neurotol 2008; 29(2):189-92.
Büchner A, Brendel M, Krüeger B, Frohne-Büchner C, Nogueira W, Edler B, Lenarz T. Current 26.
steering and results from novel speech coding strategies. Otol Neurotol 2008; 29(2):203-7.
Donaldson GS, Kreft HA, Litvak L. Place-pitch discrimination of single- versus dual-electrode 27.
stimuli by cochlear implant users (L). J Acoust Soc Am 2005; 118(2):623-6.
Koch DB, Downing M, Osberger MJ, Litvak L. Using current steering to increase spectral 28.
resolution in CII and HiRes 90K users. Ear Hear 2007; 28(2 Suppl):38S-41S.
Kwon BJ, van den Honert C. Dual-electrode pitch discrimination with sequential interleaved 29.
stimulation by cochlear implant users. J Acoust Soc Am 2006; 120(1):EL1-6.
Brendel M, Buechner A, Krueger B, Frohne-Buechner C, Lenarz T. Evaluation of the Harmony 30.
sound processor in combination with the speech coding strategy HiRes 120. Otol Neurotol 2008; 29(2):199-202.
Saoji AA, Litvak L. Use of virtual electrodes generated with simultaneous stimulation to 31.
program around disabled electrodes in Fidelity120. Abstracts of the 10th international conference on cochlear implants and other implantable auditory technologies 2008; San Diego, California, 66 (poster presentation).
Saoji AA, Litvak L. Equal loudness contours and discrete pitch percepts for simultaneous 32.
dual-electrode stimulation as a function of spatial electrode separation. Abstract of the 10th international conference on cochlear implants and other implantable auditory technologies 2008; San Diego, California, 86 (oral presentation).
Saoji, A.A., Litvak, L. Excitation Patterns for Physical Electrode and Simultaneous Dual- 33.
Electrode Stimulation Measured Using Neural Response Imaging (NRI), Abstract of ARO Midwinter Meeting 2008; Phoenix, Arizona, 472 (poster presentation).
Briaire JJ, Frijns JHM. Unraveling the electrically evoked compound action potential. Hear 34.
Res 2005; 205(1-2):143-56.
Cohen LT. Practical model description of peripheral neural excitation in cochlear implant 35.
recipients: 4. Model development at low pulse rates: General model and application to individuals. Hear Res 2009; 248(1-2):15-30.
Cohen LT. Practical model description of peripheral neural excitation in cochlear implant 36.
recipients: 5. Refractory recovery and facilitation. Hear Res 2009; 248(1-2):1-14.
Brown CJ, Abbas PJ, Gantz B. Electrically evoked whole-nerve action potentials: data from 37.
human cochlear implant users. J Acoust Soc Am 1990; 88(3):1385-91.
Klop WM, Hartlooper A, Briare JJ, Frijns JH. A new method for dealing with the stimulus 38.
artefact in electrically evoked compound action potential measurements. Acta Otolaryngol 2004; 124(2):137-43.
Frijns JH, Briaire JJ, De Laat JA, Grote JJ. 2002. Initial evaluation of the Clarion CII cochlear 39.
implant: speech perception and neural response imaging. Ear Hear 2002; 23:184-197.
Frijns JH, Kalkman RK, Vanpoucke FJ, Bongers JS, Briaire JJ. Simultaneous and non- 40.
simultaneous dual electrode stimulation in cochlear implants: evidence for two neural response modalities. Acta Otolaryngol.2008; 31:1-7 (Epub ahead of print).
Rubinstein JT, Wilson BS, Finley CC. Abbas PJ. Pseudospontaneous activity: stochastic 41.
independence of auditory nerve fibers with electrical stimulation. Hear Res 1999; 127(1- 2):108-18.
Briaire JJ, Frijns JHM. The consequences of neural degeneration regarding optimal cochlear 42.
implant position in scala tympani: a model approach. Hear Res 2006; 214(1-2):17-27.
Cohen LT, Saunders E, Clark GM. Psychophysics of a prototype peri-modiolar cochlear 43.
implant electrode array. Hear Res. 2001; 155(1-2):63-81.
van der Beek FB, Boermans PP, Verbist BM, Briaire JJ, Frijns JH. Clinical evaluation of the 44.
Clarion CII HiFocus 1 with and without positioner. Ear Hear 2005; 26(6):577-92.
